Code for our NeurIPS 2021 paper "Mining the Benefits of Two-stage and One-stage HOI Detection".
Contributed by Aixi Zhang*, Yue Liao*, Si Liu, Miao Lu, Yongliang Wang, Chen Gao and Xiaobo Li.
Installl the dependencies.
pip install -r requirements.txt
HICO-DET dataset can be downloaded here. After finishing downloading, unpack the tarball (hico_20160224_det.tar.gz) to the data directory.
Instead of using the original annotations files, we use the annotation files provided by the PPDM authors. The annotation files can be downloaded from here. The downloaded annotation files have to be placed as follows.
data
└─ hico_20160224_det
|─ annotations
| |─ trainval_hico.json
| |─ test_hico.json
| └─ corre_hico.npy
:
First clone the repository of V-COCO from here, and then follow the instruction to generate the file instances_vcoco_all_2014.json. Next, download the prior file prior.pickle from here. Place the files and make directories as follows.
CDN
|─ data
│ └─ v-coco
| |─ data
| | |─ instances_vcoco_all_2014.json
| | :
| |─ prior.pickle
| |─ images
| | |─ train2014
| | | |─ COCO_train2014_000000000009.jpg
| | | :
| | └─ val2014
| | |─ COCO_val2014_000000000042.jpg
| | :
| |─ annotations
: :
For our implementation, the annotation file have to be converted to the HOIA format. The conversion can be conducted as follows.
PYTHONPATH=data/v-coco \
python convert_vcoco_annotations.py \
--load_path data/v-coco/data \
--prior_path data/v-coco/prior.pickle \
--save_path data/v-coco/annotations
Note that only Python2 can be used for this conversion because vsrl_utils.py in the v-coco repository shows a error with Python3.
V-COCO annotations with the HOIA format, corre_vcoco.npy, test_vcoco.json, and trainval_vcoco.json will be generated to annotations directory.
Download the pretrained model of DETR detector for ResNet50, and put it to the params directory.
python convert_parameters.py \
--load_path params/detr-r50-e632da11.pth \
--save_path params/detr-r50-pre-2stage-q64.pth \
--num_queries 64
python convert_parameters.py \
--load_path params/detr-r50-e632da11.pth \
--save_path params/detr-r50-pre-2stage.pth \
--dataset vcoco
After the preparation, you can start training with the following commands. The whole training is split into two steps: CDN base model training and dynamic re-weighting training. The trainings of CDN-S for HICO-DET and V-COCO are shown as follows.
python -m torch.distributed.launch \
--nproc_per_node=8 \
--use_env \
main.py \
--pretrained params/detr-r50-pre-2stage-q64.pth \
--output_dir logs \
--dataset_file hico \
--hoi_path data/hico_20160224_det \
--num_obj_classes 80 \
--num_verb_classes 117 \
--backbone resnet50 \
--num_queries 64 \
--dec_layers_hopd 3 \
--dec_layers_interaction 3 \
--epochs 90 \
--lr_drop 60 \
--use_nms_filter
python -m torch.distributed.launch \
--nproc_per_node=8 \
--use_env \
main.py \
--pretrained logs/checkpoint_last.pth \
--output_dir logs/ \
--dataset_file hico \
--hoi_path data/hico_20160224_det \
--num_obj_classes 80 \
--num_verb_classes 117 \
--backbone resnet50 \
--num_queries 64 \
--dec_layers_hopd 3 \
--dec_layers_interaction 3 \
--epochs 10 \
--freeze_mode 1 \
--obj_reweight \
--verb_reweight \
--lr 1e-5 \
--lr_backbone 1e-6 \
--use_nms_filter
python -m torch.distributed.launch \
--nproc_per_node=8 \
--use_env \
main.py \
--pretrained params/detr-r50-pre-2stage.pth \
--output_dir logs \
--dataset_file vcoco \
--hoi_path data/v-coco \
--num_obj_classes 81 \
--num_verb_classes 29 \
--backbone resnet50 \
--num_queries 100 \
--dec_layers_hopd 3 \
--dec_layers_interaction 3 \
--epochs 90 \
--lr_drop 60 \
--use_nms_filter
python -m torch.distributed.launch \
--nproc_per_node=8 \
--use_env \
main.py \
--pretrained logs/checkpoint_last.pth \
--output_dir logs/ \
--dataset_file vcoco \
--hoi_path data/v-coco \
--num_obj_classes 81 \
--num_verb_classes 29 \
--backbone resnet50 \
--num_queries 100 \
--dec_layers_hopd 3 \
--dec_layers_interaction 3 \
--epochs 10 \
--freeze_mode 1 \
--verb_reweight \
--lr 1e-5 \
--lr_backbone 1e-6 \
--use_nms_filter
You can conduct the evaluation with trained parameters for HICO-DET as follows.
python -m torch.distributed.launch \
--nproc_per_node=8 \
--use_env \
main.py \
--pretrained pretrained/hico_cdn_s.pth \
--dataset_file hico \
--hoi_path data/hico_20160224_det \
--num_obj_classes 80 \
--num_verb_classes 117 \
--backbone resnet50 \
--num_queries 64 \
--dec_layers_hopd 3 \
--dec_layers_interaction 3 \
--eval \
--use_nms_filter
Firstly, you need the add the following main function to the vsrl_eval.py in data/v-coco.
if __name__ == '__main__':
import sys
vsrl_annot_file = 'data/vcoco/vcoco_test.json'
coco_file = 'data/instances_vcoco_all_2014.json'
split_file = 'data/splits/vcoco_test.ids'
vcocoeval = VCOCOeval(vsrl_annot_file, coco_file, split_file)
det_file = sys.argv[1]
vcocoeval._do_eval(det_file, ovr_thresh=0.5)
Next, for the official evaluation of V-COCO, a pickle file of detection results have to be generated. You can generate the file with the following command. and then evaluate it as follows.
python generate_vcoco_official.py \
--param_path pretrained/vcoco_cdn_s.pth \
--save_path vcoco.pickle \
--hoi_path data/v-coco \
--dec_layers_hopd 3 \
--dec_layers_interaction 3 \
--use_nms_filter
cd data/v-coco
python vsrl_eval.py vcoco.pickle
| Full (D) | Rare (D) | Non-rare (D) | Full(KO) | Rare (KO) | Non-rare (KO) | Download | |
|---|---|---|---|---|---|---|---|
| CDN-S (R50) | 31.44 | 27.39 | 32.64 | 34.09 | 29.63 | 35.42 | model |
| CDN-B (R50) | 31.78 | 27.55 | 33.05 | 34.53 | 29.73 | 35.96 | model |
| CDN-L (R101) | 32.07 | 27.19 | 33.53 | 34.79 | 29.48 | 36.38 | model |
D: Default, KO: Known object
| Scenario 1 | Scenario 2 | Download | |
|---|---|---|---|
| CDN-S (R50) | 61.68 | 63.77 | model |
| CDN-B (R50) | 62.29 | 64.42 | model |
| CDN-L (R101) | 63.91 | 65.89 | model |
Please consider citing our paper if it helps your research.
@article{zhang2021mining,
title={Mining the Benefits of Two-stage and One-stage HOI Detection},
author={Zhang, Aixi and Liao, Yue and Liu, Si and Lu, Miao and Wang, Yongliang and Gao, Chen and Li, Xiaobo},
journal={arXiv preprint arXiv:2108.05077},
year={2021}
}
CDN is released under the Apache 2.0 license. See LICENSE for additional details.
Some of the codes are built upon PPDM, DETR and QPIC. Thanks them for their great works!
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent